Similar Face Detection for Indian Faces using Siamese Neural Networks

摘要

Now a days facial recognition is a very common yet highly significant task all around the globe. Most of the times facial detection or facial similarity detection would lead the thought towards complex neural networks like CNN’s. But the task of working with CNN’s would be very computationally expensive and also require huge amounts of data for better results. In this work we focus on implementing the task of facial similarity detection using Siamese neural networks, a rather simple architecture which is capable of one-shot learning. A new dataset has been created and tested for facial similarity using Siamese neural networks and the model is able to predict the top similar images of a given image very accurately.